MRG: update to plot_2d fixes and tests

.travis.yml

@@ -45,6 +45,8 @@ script:
     # Change into an innocuous directory and find tests from installation
     - mkdir for_testing
     - cd for_testing
+    # No figure windows for mpl; quote to hide : from travis-ci yaml parsing
+    - 'echo "backend : agg" > matplotlibrc'
     - if [ "${COVERAGE}" == "1" ]; then
       cp ../.coveragerc .;
       COVER_ARGS="--with-coverage --cover-package dipy";

dipy/align/imwarp.py

@@ -611,16 +611,18 @@ def _warp_forward(self, image, interpolation='linear', world_to_image=None,
         space we need to bring via interpolation). So, S is the matrix that
         converts the sampling grid (whose shape is given as parameter
         'sampling_shape' ) to space coordinates.
-
         """
         #if no world-to-image transform is provided, we use the codomain info
         if world_to_image is None:
             world_to_image = self.codomain_affine_inv
         #if no sampling info is provided, we use the domain info
         if sampling_shape is None:
             if self.domain_shape is None:
-                raise ValueError('Unable to infer sampling info. Provide a valid sampling_shape.')
+                raise ValueError('Unable to infer sampling info. '
+                                 'Provide a valid sampling_shape.')
             sampling_shape = self.domain_shape
+        else:
+            sampling_shape = np.asarray(sampling_shape, dtype=np.int32)
         if sampling_affine is None:
             sampling_affine = self.domain_affine
 
@@ -801,12 +803,14 @@ def transform(self, image, interpolation='linear', world_to_image=None,
         See _warp_forward and _warp_backward documentation for further
         information.
         """
+        if sampling_shape is not None:
+            sampling_shape = np.asarray(sampling_shape, dtype=np.int32)
         if self.is_inverse:
             warped = self._warp_backward(image, interpolation, world_to_image,
-                                       sampling_shape, sampling_affine)
+                                         sampling_shape, sampling_affine)
         else:
             warped = self._warp_forward(image, interpolation, world_to_image,
-                                       sampling_shape, sampling_affine)
+                                        sampling_shape, sampling_affine)
         return np.asarray(warped)
 
     def transform_inverse(self, image, interpolation='linear', world_to_image=None,